Rational Synthesis of Low-Polydispersity Block Copolymer Vesicles inConcentrated Solution via Polymerization-Induced Self-AssemblyCarlo Gonzato,† Mona Semsarilar,† Elizabeth R. Jones,† Feng Li,‡ Gerard J. P. Krooshof,‡ Paul Wyman,§

Oleksandr O. Mykhaylyk,*,† Remco Tuinier,‡,∥ and Steven P. Armes*,†

†Department of Chemistry, The University of Sheffield, Brook Hill, Sheffield, S3 7HF, United Kingdom‡DSM ChemTech Center, Advanced Chemical Engineering Solutions (ACES), P.O. Box 18, 6160 MD Geleen, The Netherlands§DSM Ahead, P.O. Box 18, 6160 MD Geleen, The Netherlands∥Van ’t Hoff Laboratory for Physical and Colloid Chemistry, Department of Chemistry, Utrecht University, Padualaan 8, 3584 CHUtrecht, The Netherlands

*S Supporting Information

ABSTRACT: Block copolymer self-assembly is normallyconducted via post-polymerization processing at high dilution.In the case of block copolymer vesicles (or “polymersomes”),this approach normally leads to relatively broad sizedistributions, which is problematic for many potentialapplications. Herein we report the rational synthesis of low-polydispersity diblock copolymer vesicles in concentratedsolution via polymerization-induced self-assembly using reversible addition−fragmentation chain transfer (RAFT) polymer-ization of benzyl methacrylate. Our strategy utilizes a binary mixture of a relatively long and a relatively short poly(methacrylicacid) stabilizer block, which become preferentially expressed at the outer and inner poly(benzyl methacrylate) membrane surface,respectively. Dynamic light scattering was utilized to construct phase diagrams to identify suitable conditions for the synthesis ofrelatively small, low-polydispersity vesicles. Small-angle X-ray scattering (SAXS) was used to verify that this binary mixtureapproach produced vesicles with significantly narrower size distributions compared to conventional vesicles prepared using asingle (short) stabilizer block. Calculations performed using self-consistent mean field theory (SCMFT) account for thepreferred self-assembled structures of the block copolymer binary mixtures and are in reasonable agreement with experiment.Finally, both SAXS and SCMFT indicate a significant degree of solvent plasticization for the membrane-forming poly(benzylmethacrylate) chains.

■ INTRODUCTION

The self-assembly of AB diblock copolymers in selective solventsis a well-known approach for generating a wide range ofcopolymer morphologies, including spherical micelles, worm-like micelles, or vesicles.1,2 The latter morphology is of particularinterest, since it can be exploited for drug delivery, encapsulation,cosmetics, biomimetic applications, and as nanoreactors.3−7

Compared with liposomes,8 block copolymer vesicles exhibitgreater toughness, reduced permeability, and less fluid-likemembranes.3,7,9

Amphiphilic diblock copolymer nano-objects are usuallygenerated via postpolymerization processing using either asolvent switch, pH switch, dialysis, or thin film rehydration; suchprocessing is typically conducted in (mixed) aqueous solution atlow copolymer concentrations (typically <1%).10 The resultingcopolymers are often in their frozen, non-ergodic state,10 inwhich case the morphologies that are observed are path-dependent. This can be an advantage, because kineticallycontrolled self-assembly can allow access to various nano-objectsfrom relatively few copolymer compositions. However, thisapproach can become problematic when a high degree of controlover the particle size distribution is desired. This is a particular

problem for vesicles, which form via “wrap-up” of planarmembranes in solution.11 Sonication,12 membrane extrusion,13

high-pressure homogenization,14 self-assembly onto a patternedtemplate,15 or thermal cycling13 have been used to gain controlover vesicle size distributions. In addition, inkjet printing has alsobeen demonstrated to be effective in controlling vesicle size.16 Inthe above examples, narrower size distributions are typicallyachieved, with a concomitant reduction in vesicle dimensions.This is because the interfacial area (and hence interfacial energy)for smaller vesicles is much more size-dependent than for largervesicles.17 In addition to such physical processing routes,chemical strategies have also been explored for producing low-polydispersity vesicles. For example, Eisenberg and co-workersreported that, for poly(acrylic acid)-polystyrene (PAA-PS)diblocks, utilizing an artificially broad distribution of PAAblock lengths in combination with a fixed degree of polymer-ization (DP) of the membrane-forming PS block aided theformation of relatively small, low-polydispersity vesicles.18,19

This was attributed to self-segregation of the PAA chains: longer

Received: May 29, 2014Published: July 15, 2014

Article

pubs.acs.org/JACS

© 2014 American Chemical Society 11100 dx.doi.org/10.1021/ja505406s | J. Am. Chem. Soc. 2014, 136, 11100−11106

chains were preferentially located at the external membranesurface, whereas shorter chains were expressed at the innermembrane surface. This hypothesis was supported by an elegantfluorescence labeling experiment.19 The feasibility of exertinglocal thermodynamic control over systems that are globally non-ergodic has been extensively studied both theoretically andexperimentally for blends of diblock copolymer amphi-philes20−26 as well as for asymmetric triblock or tetrablockcopolymers.4,27 Although enhanced control over the vesicularmorphology can be achieved using this approach, its multistepnature and the relatively low copolymer concentrations utilizedto date constitute major drawbacks with regard to a cost-effectivescalable route. However, the recent development of robustpolymerization-induced self-assembly (PISA) formulationsbased on living radical polymerization offers an opportunity toovercome such limitations: vesicles can now be prepared directlyin solution at up to 25% solids using either aqueous emulsion28,29

or alcoholic30,31 dispersion polymerization. In the present work,we report the rational synthesis of low-polydispersity diblockcopolymer vesicles via PISA using reversible addition−fragmentation chain transfer (RAFT) dispersion polymerizationin concentrated ethanolic solution. In particular, binary mixturesof two poly(methacrylic acid) (PMAA) macromolecular chaintransfer agents (macro-CTAs) are chain-extended with benzylmethacrylate (BzMA), see Scheme 1. The resulting diblockcopolymer vesicles are characterized using transmission electronmicroscopy (TEM), dynamic light scattering (DLS), and small-angle X-ray scattering (SAXS). For comparison, PISA synthesesusing the individual PMAA macro-CTAs were also performed ascontrol experiments. Finally, we analyzed the thermodynamicallypreferred self-assembled copolymer vesicle morphologies usingself-consistent mean-field theory (SCMFT).

■ MATERIALS AND METHODSAll reagents were purchased from Sigma-Aldrich (U.K.) and used asreceived unless otherwise stated. 4,4′-Azobis-4-cyanopentanoic acid(ACVA, >98%) was used as an initiator. Benzyl methacrylate (BzMA,96%) was passed through a column (supplied by the manufacturer) inorder to remove inhibitor prior to use. 4-Cyano-4-(2-phenylethane-sulfanylthiocarbonyl)sulfanylpentanoic acid (PETTC) was synthesizedas previously reported.30

Synthesis of PMAA macro-CTAs. For the typical synthesis of aPMAA62 macro-CTA, a round-bottomed flask was charged with MAA(10.0 g; 116 mmol), PETTC (0.607 g; 1.79 mmol), ACVA (0.100 g;0.36 mmol; PETTC/ACVA molar ratio = 5.0), and ethanol (10.0 g).The sealed flask was cooled using an ice bath and purged with nitrogengas, then placed in a preheated oil bath at 70 °C for 3 h. The resultingPMAA macro-CTA (MAA conversion =99%; after exhaustivemethylation using trimethylsilyl diazomethane, Mn = 6800 g mol−1,Mw = 8800 g mol

−1,Mw/Mn = 1.29), was purified by dialysis, first against9:1 water/methanol and then against deionized water. The polymer wasisolated via lyophilization. A mean DP of 62 was calculated for thismacro-CTA using 1H NMR spectroscopy by comparing the integratedsignal intensity assigned to the aromatic protons at 7.2−7.4 ppm withthat due to the methacrylic polymer backbone at 0.4−2.5 ppm. Usingthis same general protocol, PMAA44, PMAA102, and PMAA171 weresynthesized by adjusting the amounts of PETTC and ACVA utilized toachieve the desired target DP (see Table S1 for further details).

Synthesis of poly(methacrylic acid)-poly(benzyl methacry-late) (PMAA-PBzMA) diblock copolymer vesicles by dispersionpolymerization in ethanol. In a typical dispersion polymerizationtargeting PMAA44-PBzMA140 at 20% w/w solids, BzMA (0.40 g; 2.27mmol), ACVA (0.90 mg; 0.003 mmol), and PMAA44 (0.067 g; 0.016mmol) were dissolved in ethanol (1.871 g). This reaction mixture wasice-cooled, nitrogen-purged, and placed in a preheated oil bath at 70 °Cfor 24 h. The monomer conversion was determined by 1H NMRspectroscopy, by comparing the integrated benzyl signal assigned toPBzMA at 4.9 ppm to that of the BzMA vinyl proton signal (CH2) at 5.2ppm.

Synthesis of low-polydispersity diblock copolymer vesiclesvia dispersion polymerization of BzMA in ethanol using abinary mixture of two PMAA macro-CTAs. In a typical dispersionpolymerization targeting (PMAA44 + PMAA102)-PBzMA140 at 20% w/wsolids, BzMA (0.40 g; 2.27 mmol), ACVA (0.90 mg; 0.003 mmol),PMAA44 (0.033 g; 0.008 mmol), and PMAA102 (0.074 g; 0.008 mmol)were dissolved in ethanol (2.033 g). The reaction mixture was ice-cooled, purged using nitrogen gas, and placed in a preheated oil bath at70 °C for 24 h. The polymerization yield was determined by 1H NMRspectroscopy, by comparing the integrated benzyl proton signal assignedto PBzMA at 4.9 ppm to that of the BzMA vinyl signal (CH2) at 5.2 ppm.

1HNMR Spectroscopy. 1HNMR spectra were acquired in CD3OD,CD2Cl2, CDCl3 or d6-DMSO using a Bruker 400 MHz spectrometer,with chemical shifts reported in ppm.

Transmission Electron Microscopy (TEM). TEM studies wereconducted using a Philips CM 100 instrument operating at 100 kV.TEM samples were prepared as follow: copper-palladium TEM grids(Agar Scientific, U.K.) were surface-coated in-house to yield a thin film

Scheme 1. PISA Synthesis of Diblock Copolymer Vesicles via RAFT Dispersion Polymerization of BzMA in Ethanol at 70°Ca

a(a) chain extension of a single PMAA macro-CTA to give high-polydispersity vesicles; (b) chain extension of two PMAA macro-CTAs having a low(green) and a high (red) mean degree of polymerization, respectively, to produce relatively low-polydispersity vesicles.

Journal of the American Chemical Society Article

dx.doi.org/10.1021/ja505406s | J. Am. Chem. Soc. 2014, 136, 11100−1110611101

of amorphous carbon. The grids were then plasma glow-discharged for35 s in order to create a hydrophilic surface. Ethanolic dispersions(0.20% w/w, 10 μL) were then placed onto a freshly glow-dischargedgrid for 1 min and then blotted with filter paper to remove excesssolution. To stain the deposited particles, a 0.75% w/w aqueous solutionof uranyl formate (10 μL) was placed via micropipette on a sample-loaded grid for 20 s and then carefully blotted to remove the excess stain.Each grid was then carefully dried using a vacuum hose.Gel Permeation Chromatography (GPC). The GPC set-up

comprised two 5 μm (30 cm) Mixed C columns and a WellChrom K-2301 refractive index detector operating at 950 ± 30 nm. THF eluentcontaining 2.0% v/v triethylamine and 0.05% w/v butylhydroxytoluene(BHT) was used at a flow rate of 1.0 mL min−1. A series of ten near-monodisperse linear poly(methyl methacrylate) standards (Mp rangingfrom 1280 to 330,000 g mol−1) were purchased from Agilent (ChurchStretton, U.K.) and employed for calibration using the above refractiveindex detector. Both PMAA macro-CTAs and PMAA-PBzMA diblocksrequired exhaustive methylation of the methacrylic acid units usingexcess trimethylsilyldiazomethane.32

Dynamic Light Scattering (DLS). DLS measurements wereconducted using a Malvern Instruments Zetasizer Nano-ZS instrumentequipped with a 4 mW He-Ne laser operating at 633 nm, an avalanchephotodiode detector at a fixed angle of 173°, and an ALV/LSE-5003multiple τ digital correlator electronics system. Autocorrelationfunctions were analyzed by the cumulants method to calculate the z-average (hydrodynamic) diameter and polydispersity index (PDI).Small Angle X-ray Scattering (SAXS). SAXS patterns were

recorded at two synchrotron sources (ESRF, station ID02, Grenoble,France and Diamond Light Source, station I22, Didcot, U.K.) usingmonochromatic X-ray radiation (wavelength λ = 0.0995 and 0.1001 nm,respectively) and 2D CCD detectors (FReLoN Kodak and Pilatus 2M,respectively). The camera length set-ups covered q ranges from 0.01 to1.9 nm−1 and from 0.02 to 1.9 nm−1, respectively, where q = 4π sin θ/λ isthe modulus of the scattering vector and θ is half of the scattering angle.A liquid cell comprising two mica windows (each of 25 μm thickness)separated by a polytetrafluoroethylene spacer of 1 mm thickness wasused as a sample holder. Scattering data reduced by Nika SAS datareduction macros for Igor Pro (integration, normalization andbackground subtraction) were further analyzed using Irena SAS macrosfor Igor Pro.33 The SAXS measurements were conducted on selectedethanolic PMAA-PBzMA copolymer dispersions diluted from the as-synthesized concentrated dispersions to 1.0% w/w solids (Table 1).

■ RESULTS AND DISCUSSION

Our synthetic strategy is based on the simultaneous chainextension of two hydrophilic macro-CTAs comprising relativelyshort and long DPs (e.g., PMAA62 and PMAA171) with ahydrophobic monomer (i.e., BzMA). The GPC traces shown inFigures S1−S3 confirm that the binary mixture PISA formulationis essentially equivalent to the conventional PISA formulationbased on the sole PMAA62 macro-CTA. In both cases, well-defined diblock copolymer chains with relatively low Mw/Mn

values and high blocking efficiencies are obtained, with minimalGPC evidence for macro-CTA contamination. However, usingthe binary mixture of the two macro-CTAs understandablyproduces a slightly higher copolymer polydispersity (e.g., anMw/Mn of 1.36 in Figure S1). Using such a binary mixture of macro-CTAs, two amphiphilic diblock copolymers are generated in situthat stabilize differing particle morphologies:34 PMAA62-PBzMAx favors vesicular self-assembly, while PMAA171-PBzMAxprefers to form spherical micelles (where x is the target DP of themembrane-forming and core-forming PBzMA block respec-tively; in these experiments x is such that 180 ≤ x ≤ 350).According to the literature on the self-assembly of binarymixtures of block copolymers,18,21,24 the longer PMAA171 blockshould be preferentially located on the outer membrane surface,while the shorter PMAA62 block should be expressed at the innersurface. Unlike post-polymerization processing, PISA formula-tions involve efficient self-assembly at relatively high copolymerconcentration, which is a potentially decisive advantage forindustrial scale-up. Binary mixtures of chemically differentmacro-CTAs have been recently explored for PISA formula-tions.35,36 However, as far as we are aware the present work is thefirst to exploit PISA to target relatively small, low-polydispersityvesicles.Initially, we synthesized a series of PMAA62-PBzMAx

(targeting x = 180−350) vesicles at 20% w/w solids in ethanol.DLS and TEM characterization (see Figure S4 and Table S2)confirmed that these vesicles were invariably relatively large andpolydisperse, and hence served as a useful reference point (N.B.the PBzMA Tg is ∼55 °C, so these vesicles tend to resist collapseunder high vacuum). Conversely, small vesicles tended to haverelatively narrow size distributions, since the interfacial area (andhence energy) is strongly size dependent.17 Thus, reducing thevesicle dimensions should simultaneously lower their poly-dispersity. This hypothesis was examined using DLS, whichreports the z-average diameter and polydispersity, to construct aphase diagram (see Figure 1; target PBzMA DP = 350) usingvarious binary mixtures of PMAA62 and PMAA171 macro-CTAsfor the RAFT dispersion polymerization of BzMA in ethanol.Systematic adjustment of the PMAA62 mole fraction resulted incopolymer morphologies ranging from small spherical nano-particles to large vesicles (see Figure 1). Empirically, a PMAA62mole fraction of around 0.50−0.60 offered optimal morpho-logical control. At lower mole fractions, the influence of themajor component (i.e., the longer PMAA171 block) leads to theformation of spherical nanoparticles, while at relatively highPMAA62 mole fractions only rather large, polydisperse vesiclesare produced. Moreover, there is a significant reduction in vesiclepolydispersity at intermediate PMAA62 mole fractions, as judged

Table 1. SAXS and TEM Data for (0.40 PMAA62 + 0.60 PMAA171)-PBzMA312, (0.50 PMAA62 + 0.50 PMAA171)-PBzMA326, andPMAA62-PBzMA333 diblock copolymers synthesized by RAFTDispersion Polymerization of BzMA at 70°CUsing a BinaryMixtureof PMAA62 and PMAA171 macro-CTAs in Ethanol and 20% w/w Solidsa

diameter, nm PDIc membrane thickness, nm

copolymer composition copolymer morphology target DP actual DPb SAXS SAXS TEMd SAXS

(0.4 M62 + 0.6 M171)-B312 micelles 350 312 62 ± 2.5 0.01 − −(0.5 M62 + 0.5 M171)-B326 vesicles 350 326 199 ± 20 0.05 31 ± 3 29 ± 4M62-B333 vesicles 350 333 351 ± 60 0.12 ∼50 38 ± 4

aHerein ‘M’ indicates PMAA, while ‘B’ denotes PBzMA. bMean DPs for the core-forming PBzMA (or B block) were determined by 1H NMRspectroscopy; cPolydispersities (PDI) were calculated as the square of the standard deviation of the diameter divided by the diameter; dImageJsoftware was used to estimate membrane thicknesses, with data being averaged over 40 vesicles. The mean membrane thickness of the (0.50 PMAA62+ 0.50 PMAA171)-PBzMA326 vesicles was determined from cryo-TEM images (see Figure S6), while conventional TEM images were used for thePMAA62-PBzMA333 vesicles.

Journal of the American Chemical Society Article

dx.doi.org/10.1021/ja505406s | J. Am. Chem. Soc. 2014, 136, 11100−1110611102

by DLS (see Figure 1). Compared to the reference PMAA62-PBzMA350 vesicles (z-average diameter = 405 nm, DLSpolydispersity = 0.42), the (xPMAA62 + (1 − x)PMAA171)-PBzMA350 vesicles formed at x = 0.50 have a significantly lower z-average diameter of 188 nm and a substantially reducedpolydispersity of 0.02. These observations are consistent withearlier experimental studies that involved post-polymerizationprocessing of diblock copolymers in dilute aqueous solution,18

and also with theoretical predictions made by Greenall et al. forthe self-assembly of binary mixtures of sphere-forming diblockcopolymers with vesicle-forming diblock copolymers.24 SAXSwas utilized to gain further physical insight regarding themorphological transition for the [xPMAA62 + (1− x)PMAA171]-PBzMA350 formulation on moving from x = 0.40 to 0.50. It wasfound that either a vesicle model or a spherical micelle model(see Supporting Information for descriptions of these twomodels) produced reasonably good fits to the appropriate SAXSpatterns over five orders of magnitude of X-ray scatteringintensity (Figure 2). A scattering pattern obtained for the (0.4PMAA62 + 0.6 PMAA171)-PBzMA312 formulation, with theintensity curve tending toward a zero gradient for q < 0.1 nm−1,corresponds to a spherical micelle morphology (see SAXS data inFigure 2a). In contrast, scattering patterns displaying acontinuous increase in intensity at q < 0.1 nm−1 are observedfor copolymer dispersions containing a relatively high proportion

of the short PMAA62 block (see SAXS data in Figures 2b,c). Boththese curves can be satisfactorily fitted using the vesicle model.Thus, SAXS data recorded for the ethanolic copolymerdispersions are consistent with TEM studies of the driedcopolymer nanoparticles (Figure 2): a sphere-to-vesiclemorphological transition occurs between the (0.40 PMAA62 +0.60 PMAA171)-PBzMA312 and (0.50 PMAA62 + 0.50 PMAA171)-PBzMA326 binary mixture formulations (Figure 1). For thespherical micelle model, it is assumed that the corona is formedby a mixture of uniformly distributed PMAA62 and PMAA171

blocks, thus an average radius of gyration was used for the modelfitting (Table S4). According to Eisenberg and co-workers,18,19

for vesicles formed by a binary mixture of diblock copolymerscomprising essentially the same mean DP for the membrane-forming block, the shorter stabilizer block should be preferen-tially expressed at the inner leaflet and the longer stabilizer blockshould be located mainly at the outer leaflet. Such segregationenables the differing packing parameters for the two copolymerchains to be accommodated within the same nano-object.Accordingly, a reasonable approximation that accounts for thislocalized copolymer segregation is included in the vesicle modelused to analyze the SAXS data. Thus two different radii ofgyration for the PMAA62 and PMAA171 blocks (Rg in and Rg out,respectively, see Supporting Information) were used in themodel. An averaged distribution of these blocks across the

Figure 1. Representative TEM images and corresponding phase diagram determined for [xPMAA62 + (1 − x)PMAA171]-PBzMA350 particlessynthesized by RAFT dispersion polymerization of BzMA at 70 °C using a binary mixture of PMAA62 and PMAA171 macro-CTAs in ethanol at 20% w/wsolids. Intensity-average diameters and polydispersities were determined by DLS in ethanol (0.10% w/w solids). S indicates spheres and V indicatesvesicles.

Journal of the American Chemical Society Article

dx.doi.org/10.1021/ja505406s | J. Am. Chem. Soc. 2014, 136, 11100−1110611103

membrane was also considered, but the former approach gave aslightly better fit (see SAXS data in Figure 2b and Table S4). TheSAXS pattern obtained for the high-polydispersity PMAA62-PBzMA333 vesicles was fitted using a simplified model such thatRg in = Rg out (see SAXS data in Figure 2c and Table S4). SAXSanalysis indicates that the average radius of gyration of the PMAAstabilizer blocks is reduced from 3.9 nm for the spherical micellesto 2.9 nm for the polydisperse vesicles formed by the PMAA62-PBzMA333 alone (Table S4). This is physically reasonable, sincethe coronal layer of the polydisperse vesicles solely comprisesPMAA62 chains, whereas the spherical micelles prepared usingthe binary mixture of macro-CTAs necessarily contains asignificant proportion of the longer PMAA171 corona-formingblock.In principle, the Rg of the PMAA stabilizer block can be

calculated. For example, given that the projected contour lengthof a single repeat unit in the PMAA block is 0.255 nm (twocarbon bonds in an all-trans conformation), the total contourlength of the PMAA62 block, L = 62 × 0.255 nm =15.83 nm.Assuming a Kuhn length of 1.53 nm [the literature value forpoly(methyl methacrylate)37] results in an estimated Rg of (15.83× 1.53/6)0.5, or 2.01 nm. Similarly, the estimated Rg for thePMAA171 block is 3.34 nm. However, the Rg values obtained fromthe SAXS data fits are slightly higher than those calculated. This isreasonable, because the latter values indicate the unperturbeddimensions of the PMAA chains, whereas the experimental datawere acquired under better-than-theta solvent (i.e., goodsolvent) conditions. However, a second possible explanationfor the difference between the calculated and fitted Rg value is theuncertainty in the SAXS analysis. A positive correlation betweenRg and the volume fraction of ethanol within the membrane (xsol)

was observed during such data fitting. Thus reduction of xsol inthe model should result in a concomitant reduction in the Rgvalue.Close inspection of the SAXS patterns reveals further

structural information. Compared to the PMAA62-PBzMA333vesicles, the (0.5 PMAA62 + 0.5 PMAA171)-PBzMA326 binarymixture composition exhibits pronounced oscillations in the lowq region, which is associated with the overall particle size(compare q ∼ 0.02 nm−1 in Figure 2c, SAXS with q ∼ 0.05 nm−1

in Figure 2b, SAXS). This observation is consistent with DLSdata obtained for the same two copolymer dispersions, whichindicates that the (0.5 PMAA62 + 0.5 PMAA171)-PBzMA326binary mixture composition forms vesicles with a significantlylower polydispersity. The particle diameters [calculated for thespherical micelles using Ds = 2(Rs + 2Rg) and for the vesiclesusingDv = 2(Rmc + 1/2Tmc + 2Rg out), see Table S4 for definitionsof these parameters] and the vesicle membrane thicknessescorrelate reasonably well with the TEM data (see Table 1) andalso the DLS results. According to the limited data set shown inTable 1 (see entries 2 and 3), SAXS analyses also suggest that anincrease in the relative proportion of PMAA62 in the binarymixture formulation results in a simultaneous increase in boththe overall vesicle diameter and the vesicle membrane thickness.In particular, SAXS confirms that using an optimized binarymixture of PMAA62 and PMAA171 macro-CTAs enables aremarkable reduction in size and polydispersity to be achieved forPMAA-PBzMA block copolymer vesicles. Since this PISAsynthesis is conducted at relatively high solids, it represents thefirst truly scalable route to provide low-polydispersity vesicles.In order to support the above experimental studies, we also

performed numerical calculations using self-consistent mean-field theory (SCMFT).37 Over the last two decades SCMFT hasbecome a powerful technique for studying inhomogeneouspolymeric systems at equilibrium.37−39 In particular, it enablesthermodynamically preferred self-assembled structures and theirassociated physical properties to be predicted.37,38 The centralparameter in SCMFT is the mean-field free energy F. In thepresent study, F incorporates the mixing entropy, the Flory−Huggins mixing enthalpy, the local mean-field interactions(accounting for gradients) and a compressibility constraint.20

After minimizing this free energy using a self-consistent iterativeprotocol that does not require information about the equilibriummorphologies, various thermodynamic properties of theresulting equilibrium system can be calculated. In order toperform numerical SCMFT calculations, the block copolymerwas subjected to coarse-grain modeling. The three monomericrepeat units are shown in Figure S5, along with other pertinentdetails. This choice implies that each lattice unit corresponds toapproximately 0.5 nm. The interaction energies between the fourcomponents were parametrized using χ-parameters in a Flory−Huggins type formalism. These χ-parameters were determinedby COSMO-SAC, a thermodynamic model based on quantummechanical data.40 The resulting χ-parameters calculated for theRAFT polymerization temperature of 70 °C are listed in TableS5. As expected, the interactions between the benzene group andall other components are rather repulsive. For various conditions,the SCMFT grand potential was computed for differing latticegeometries (flat, cylindrical, or spherical). Suitable conditionswere identified for the formation of thermodynamically stableblock copolymer vesicles, as observed experimentally. In terms ofequilibrium structures, it was verified via SCMFT that PMAA171-PBzMA350 forms spherical micelles in ethanol at 70 °C, whilePMAA62-PBzMA350 forms vesicles under the same conditions.

Figure 2. (Upper) Representative TEM images for (a) (0.4 PMAA62 +0.6 PMAA171)-PBzMA312, (b) (0.5 PMAA62 + 0.5 PMAA171)-PBzMA326,and (c) PMAA62-PBzMA333 dried copolymer dispersions. (Lower)SAXS patterns obtained for 1.0% w/w copolymer dispersions of: (a)(0.4 PMAA62 + 0.6 PMAA171)-PBzMA312 (blue circles); (b) (0.5PMAA62 + 0.5 PMAA171)-PBzMA326 (green circles); (c) PMAA62-PBzMA333 (red circles) in ethanol. Solid lines show fits to the data usinga spherical micelle model (a) and a vesicle model (b and c).

Journal of the American Chemical Society Article

dx.doi.org/10.1021/ja505406s | J. Am. Chem. Soc. 2014, 136, 11100−1110611104

For the computed equilibrium conditions, we calculated andanalyzed the volume fraction profiles φ(r), with examples beingpresented in Figure 3 (r is the distance from the center of a

vesicle, in lattice units). The results are plotted for the densityprofiles of all components for the vesicle-forming PMAA62-PBzMA350 diblock copolymer at a copolymer concentration of 20vol % (which is close to 20% w/w) in Figure 3a and also for thebinary mixture formulation comprising 10 vol % PMAA62-PBzMA350 and 10 vol % PMAA171-PBzMA350 in Figure 3b. Inboth cases a vesicle bilayer is formed, and the profiles in Figure 3aare in agreement with the cartoon structure shown in Scheme 1a.In the case of the binary mixture formulation, the distribution ofPMAA62 chains between the inner and outer leaflets becomesmore asymmetric; the PMAA62 blocks tend to becomesegregated within the vesicle lumen (Figure 3b), as suggestedby SAXS model fitting (Figure 2b, SAXS). The differing packingparameters for the PMAA62- and PMAA171-stabilized diblockcopolymers are the driving force that dictates the preferredbilayer curvature.41,42

Furthermore, the thickness of the external vesicle coronabecomes much thicker for the binary mixture formulation,because it is the relatively long PMAA171 block that is expressedat the outer vesicle surface and hence mainly protrudes into thebulk solution. Although the binary mixture formulation for low-

polydispersity vesicles has a thicker hydrophilic shell, thehydrophobic region layer is slightly thinner (by just a few nm).This is corroborated by the TEM and SAXS data summarized inTable 1, although this effect is somewhat larger for theexperimental system. It is noteworthy that SCMFT indicates asimilar mean thickness (∼ 60 lattice units × 0.5 nm per latticeunit = ∼ 30 nm) for the membrane of the low-polydispersityvesicles, see Figure 3b. These theoretical calculations also suggestthat the longer PMAA171 chains adopt a relatively extendedconformation from both the inner and outer membrane surfaces.This is understandable, because if they remained close to theinterface, this would increase the local osmotic pressure. On theother hand, both PMAA171 and PMAA62 are covalently linked tothe hydrophobic vesicle membrane. Thus the concentration ofboth stabilizer blocks should be relatively high at (or near) themembrane surface. In this case the aggregation number for thelow-polydispersity vesicles formed using the binary mixtureformulation should be lower than that for the high-polydispersityPMAA62-PBzMA350 reference vesicles. This should result in botha reduction in mean vesicle diameter and also thinning of thevesicle membrane, which is actually observed by TEM and SAXS(see Table 1).The SCMFT calculations suggest that the solvophobic block is

partially plasticized by ethanol (φ ∼ 0.3, Figure 3). Indeed, thistheoretical prediction is supported by the SAXS analysis, whichindicates that the ethanol volume fraction (xsol) is about 0.5 forboth the vesicle membranes and the spherical micelle cores (seeTable S4).Low-polydispersity vesicles have also been obtained using

several other binary mixtures of macro-CTAs (see Figures S3−S8), thus proving the general applicability of our binary mixturePISA approach. For example, representative TEM images, DLSdata and the corresponding phase diagrams are shown in TablesS6−S8 and Figures S7−S12 for binary mixtures of PMAA44 +PMAA102 and PMAA62 + PMAA102 macro-CTAs. Closeinspection of the DLS polydispersities for various binarymixtures of PMAA macro-CTAs (see Tables S6−S8) indicatesan important trend: larger differences between the mean DPs ofthe macro-CTAs require a higher proportion of the shortermacro-CTA to achieve the minimum vesicle polydispersity. Low-polydispersity vesicles have also been obtained when usingchemically dissimilar stabilizer blocks, such as a relatively longPMAA macro-CTA and a relatively short PHPMA macro-CTA(data not shown). In this case, enthalpic incompatibility betweenthe two dissimilar stabilizer blocks is expected to drive evenstronger localized segregation across the vesicle membrane.Finally, DLS studies of vesicle dispersions (stored at 20% w/wsolids in ethanol prior to dilution) confirm that their particle sizeand polydispersity remains essentially unchanged after storage at20 °C for 15 months (e.g., original vesicle diameter = 168 nm,PDI = 0.039; aged vesicle diameter = 164 nm, PDI = 0.024). Thisindicates that these small low-polydispersity vesicles haveexcellent long-term colloidal stability.43

■ CONCLUSIONSIn summary, using a binary mixture of a relatively long sphere-forming macro-CTA and a relatively short vesicle-formingmacro-CTA enables the convenient synthesis of PMAA-PBzMA diblock copolymer vesicles with relatively narrow sizedistributions at high solids via polymerization-induced self-assembly using RAFT polymerization. Both SAXS studies andSCMFT calculations indicate that this is achieved by segregationof the PMAA stabilizer blocks, with the longer chains being

Figure 3. Radial density profiles of volume fraction φ(r) vs r (where r isthe distance from the center of a vesicle in lattice units, where one latticeunit ≈ 0.5 nm) obtained from SCMFT calculations performed forPMAA-PBzMA diblock copolymer chains in ethanol: (a) vesiclesformed by self-assembly of PMAA62-PBzMA350; (b) vesicles formed byself-assembly of a binary mixture of PMAA62-PBzMA350 (vesicle-forming) and PMAA171-PBzMA350 (sphere-forming) diblock copoly-mers.

Journal of the American Chemical Society Article

dx.doi.org/10.1021/ja505406s | J. Am. Chem. Soc. 2014, 136, 11100−1110611105

expressed at the outer leaflet and the shorter chains beingpreferentially located at the inner leaflet. This leads to relativelysmall vesicles comprising a thick outer corona and a thin innercorona, with a significant degree of solvent plasticization of themembrane-forming chains. Moreover, such copolymer segrega-tion leads to a somewhat thinner vesicle membrane compared tothat found for the relatively polydisperse vesicles obtained whenusing the shorter PMAA block as the sole stabilizer. This facileself-assembly strategy is amenable to industrial scale-up and isexpected to be of rather general utility.

■ ASSOCIATED CONTENT*S Supporting InformationSynthetic details for the PMAA-based macro-CTAs used in thisstudy, GPC curves for macro-CTAs and diblock copolymers,detailed spherical micelles and vesicle SAXS models, six furtherdiblock copolymer phase diagrams based on DLS data withcorresponding TEM images. This material is available free ofcharge via the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATION

Corresponding Authorss.p.armes@sheffield.ac.ukO.Mykhaylyk@sheffield.ac.uk

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSWe are grateful to Diamond Light Source for providingsynchrotron beam-time and thank the personnel of I22 fortheir assistance. We also acknowledge ESRF for their beam-timeallocation and wish to thank the personnel of ID02 station fortheir help with SAXS experiments described herein. We thankDSM Advanced Surfaces (Geleen, The Netherlands) forpostdoctoral support of C.G. and M.S., EPSRC for a Platformgrant to support O.O.M. (EP/J007846/1), and the ERC for anAdvanced Investigator grant for SPA (PISA 320372). The threereviewers of this manuscript are also thanked for theirconstructive criticisms.

■ REFERENCES(1) Zhang, L.; Eisenberg, A. Science 1995, 268, 1728.(2) Zhang, L.; Eisenberg, A. J. Am. Chem. Soc. 1996, 118, 3168.(3) Brinkhuis, R. P.; Rutjes, F. P.; van Hest, J. C. Polym. Chem. 2011, 2,1449.(4) Brannan, A. K.; Bates, F. S. Macromolecules 2004, 37, 8816.(5) Discher, D. E.; Ortiz, V.; Srinivas, G.; Klein, M. L.; Kim, Y.;Christian, D.; Cai, S.; Photos, P.; Ahmed, F. Prog. Polym. Sci. 2007, 32,838.(6) Choe, U.-J.; Sun, V.; Tan, J.-K.; Kamei, D. In Peptide-BasedMaterials; Deming, T., Ed.; Springer: Berlin, Heidelberg, 2012; Vol. 310,p 117.(7) LeMeins, J.-F.; Sandre, O.; Lecommandoux, S. Eur. Phys. J. E 2011,34, 1.(8) Bangham, A.; Horne, R. J. Mol. Biol. 1964, 8, 660.(9) Discher, B. M.; Won, Y.-Y.; Ege, D. S.; Lee, J. C.; Bates, F. S.;Discher, D. E.; Hammer, D. A. Science 1999, 284, 1143.(10) Hayward, R. C.; Pochan, D. J. Macromolecules 2010, 43, 3577.(11) Chen, L.; Shen, H.; Eisenberg, A. J. Phys. Chem. B 1999, 103, 9488.(12) Pereira-Lachataignerais, J.; Pons, R.; Panizza, P.; Courbin, L.;Rouch, J.; Lopez, O. Chem. Phys. Lipids 2006, 140, 88.(13) Rank, A.; Hauschild, S.; Forster, S.; Schubert, R. Langmuir 2009,25, 1337.

(14) Brandl, M.; Bachmann, D.; Drechsler, M.; Bauer, K.Drug Dev. Ind.Pharm. 1990, 16, 2167.(15) Howse, J. R.; Jones, R. A.; Battaglia, G.; Ducker, R. E.; Leggett, G.J.; Ryan, A. J. Nat. Mater. 2009, 8, 507.(16) Hauschild, S.; Lipprandt, U.; Rumplecker, A.; Borchert, U.; Rank,A.; Schubert, R.; Forster, S. Small 2005, 1, 1177.(17) Mai, Y.; Eisenberg, A. Chem. Soc. Rev. 2012, 41, 5969.(18) Terreau, O.; Luo, L.; Eisenberg, A. Langmuir 2003, 19, 5601.(19) Luo, L.; Eisenberg, A. J. Am. Chem. Soc. 2001, 123, 1012.(20) Li, F.; Marcelis, A. T. M.; Sudholter, E. J. R.; Cohen Stuart, M. A.;Leermakers, F. A. M. Soft Matter 2009, 5, 4173.(21) Li, F.; Prevost, S.; Schweins, R.; Marcelis, A. T. M.; Leermakers, F.A. M.; Cohen Stuart, M. A.; Sudholter, E. J. R. Soft Matter 2009, 5, 4169.(22) Schuetz, P.; Greenall, M. J.; Bent, J.; Furzeland, S.; Atkins, D.;Butler, M. F.; McLeish, T. C. B.; Buzza, D. M. A. Soft Matter 2011, 7,749.(23) Greenall, M. J.; Schuetz, P.; Furzeland, S.; Atkins, D.; Buzza, D.M.A.; Butler, M. F.; McLeish, T. C. B. Macromolecules 2011, 44, 5510.(24) Greenall, M. J.; Gompper, G. Macromolecules 2011, 45, 525.(25) Greenall, M. J.; Marques, C. M. Phys. Rev. Lett. 2013, 110, 088301.(26) Jain, S.; Bates, F. S. Macromolecules 2004, 37, 1511.(27) Stoenescu, R.; Meier, W. Chem. Commun. 2002, 3016.(28) Charleux, B.; Delaittre, G.; Rieger, J.; D’Agosto, F.Macromolecules2012, 45, 6753.(29) Blanazs, A.; Verber, R.; Mykhaylyk, O. O.; Ryan, A. J.; Heath, J. Z.;Douglas, C. W. I.; Armes, S. P. J. Am. Chem. Soc. 2012, 134, 9741.(30) Semsarilar, M.; Jones, E. R.; Blanazs, A.; Armes, S. P. Adv. Mater.2012, 24, 3378.(31) Sun, J.-T.; Hong, C.-Y.; Pan, C.-Y. Polym. Chem. 2013, 4, 873.(32) Graham, S.; Cormack, P. A.; Sherrington, D. C. Macromolecules2005, 38, 86.(33) Ilavsky, J.; Jemian, P. R. J. Appl. Crystallogr. 2009, 42, 347.(34) Israelachvili, J. N.; Mitchell, D. J.; Ninham, B. W. J. Chem. Soc.,Faraday Trans. 2 1976, 72, 1525.(35) Semsarilar, M.; Ladmiral, V.; Blanazs, A.; Armes, S. P. Langmuir2012, 28, 914.(36) Semsarilar, M.; Ladmiral, V.; Blanazs, A.; Armes, S. Langmuir2013, 29, 7416.(37) Fetters, L. J.; Lohsey, D. J.; Colby, R. H. In Physical Properties ofPolymers Handbook, 2nd ed.; Mark, J. E., Ed.; Springer: New York, 2007;p 447.(38) Fleer, G. J.; Cohen Stuart, M. A.; Scheutjens, J. M. H. M.;Cosgrove, B.; Vincent, B. Polymers at interfaces; Springer: New York,1993.(39) Fredrickson, G. H.; Ganesan, V.; Drolet, F.Macromolecules 2002,35, 16.(40) Claessens, M.; Van Oort, B.; Leermakers, F. A. M.; Hoekstra, F.;Cohen Stuart, M. A. Biophys. J. 2004, 87, 3882.(41)Mullins, E.; Oldland, R.; Liu, Y.; Wang, S.; Sandler, S. I.; Chen, C.-C.; Zwolak, M.; Seavey, K. C. Ind. Eng. Chem. Res. 2006, 45, 4389.(42) Lin, C.-M.; Li, C.-S.; Sheng, Y.-J.; Wu, D. T.; Tsao, H.-K.Langmuir 2012, 28, 689.(43) Leermakers, F. A. M.; Eriksson, J. C.; Lyklema, J. Associationcolloids and their equilibrium modelling, Fundamentals in Colloid andInterface Science; Elsevier: Amsterdam, 2005; Vol. 5.

Journal of the American Chemical Society Article

dx.doi.org/10.1021/ja505406s | J. Am. Chem. Soc. 2014, 136, 11100−1110611106